Plasma membrane associated transcription of cytoplasmic DNA

Abstract

Cytoplasmic membrane-associated DNA (cmDNA) is a species of DNA that attaches to the plasma membrane and has physical and chemical properties that differ from those of bulk chromosomal and mitochondrial DNAs. Here, we used deep sequencing to analyze cmDNA and showed that satellite DNAs consisting of both of simple (CCATT)_N repeats from the pericentromere regions of the chromosomes and 171-bp α-satellite repeat sequences from centromeres were highly enriched. Importantly, we found there is a special cytoplasmic membrane-associated transcription system in which DNA-dependent RNA polymerase II, which colocalizes with template cmDNA at the plasma membrane, can transcribe the membrane-associated 171-bp α-satellite repeat sequences into RNA. Analysis of phosphorylation patterns indicated that the RNA polymerase II in the plasma membrane is in a different chemical state from its nuclear counterpart.

DNA in eukaryotic organisms generally is considered to be confined to the nuclei or mitochondria of cells. However, in the early 1970s we described a species of cytoplasmic, membrane-associated DNA (cmDNA) in continuously growing lymphocytes that was associated with the plasma membrane (1, 2). Electron microscopic studies of plasma membrane fractions showed that every membrane fragment in the preparation was associated with a linear DNA fragment that appeared to make an end-on attachment to the membrane (1). This DNA appeared to originate from the nucleus, but density-labeling studies suggested that its replication was not synchronous with the rest of the genome and that after replication it exited the nucleus and became attached to cytoplasmic membranes. A study of its reassociation kinetics using the then widely used Cot_1/2 curves showed that about 73% of the cytoplasmic DNA had a rapidly reassociating component with a Cot_1/2 of about 5.0 × 10⁻³ and, in general, had a reannealing pattern that differed from bulk nuclear DNA (3–6). Since these early findings, there have been many studies of this DNA that can essentially be divided into three eras that reflect the thinking and available methodology of the times (7–10). After its initial discovery, many thought that cmDNA might be a simple contamination from nuclear DNA or some infectious agent such as a virus or mycoplasma, although none of these explanations was consistent with the analytical studies done at the time of its discovery. Then, in 1975, Saunders and coworkers (7) performed a key experiment in which they demonstrated by in situ hybridization that this DNA hybridized to the heterochromatic regions of chromosomes. These in situ studies were confirmed and extended much later by Kanda and co-workers (8), who demonstrated that cytoplasmic DNA contained α-satellite sequences that hybridized to the nucleus and the cytoplasm of cultured human cells as well as to freshly prepared bone marrow smears. This latter experiment was important because it showed that cytoplasmic DNA was present in fresh bone marrow that had not been subjected to culture conditions and that cytoplasmic DNA was not a product of cell fractionation. Although no function was associated with this DNA, Abken et al. (9) reported that cytoplasmic DNA could induce immortalization of human lymphocytes in vitro.

Today, cytoplasmic DNA is studied in the context of apoptosis, in which it is assumed to be the result of nucleic acid fragmentation in dying cells; the main analytical finding is a typical laddering effect of DNA fragments on agarose gels. Although apoptotic DNA fragments are not related directly to any physiological role for cytoplasmic DNA, these fragments have become of interest to studies of innate immunity, and it has been suggested that it might be a toll ligand, possibly related to the pathogenesis of lupus erythematosus (10). Anti-DNA autoantibodies reactive with plasma membrane-associated DNA in intact cells are used as a diagnostic marker in lupus, and it has been suggested that these antibodies represent a subset of antibodies that are different from those that are reactive with nuclear DNA (11). In summary, these studies confirm that cytoplasmic DNA originates in the genome and is present in intact cells, but whether it has any role in cellular physiology remains uncertain. Advances in nucleic acid chemistry now make it possible to characterize the nature of this DNA more precisely, and.such studies should be a prelude to any study of its role. Here, we used deep sequencing to characterize the nature of cytoplasmic DNA. In agreement with the previous hybridization results, these studies demonstrated that cmDNA is not a random sampling of the genome and that it originates largely from the centromeric and pericentromeric regions of all chromosomes from which it is removed precisely. We now show that cmDNA can be transcribed in the cytoplasm, likely because DNA-dependent RNA polymerase II colocalizes with it at the plasma membrane. Thus, we have a model in which a component of the nuclear genome exits to the cytoplasm, where it attaches to the plasma membrane and can be converted at this site into RNA. Most importantly, the present experiments suggest that, under some circumstances, the plasma membrane is a transcriptionally competent organelle.

Results

Analysis of cmDNA by Deep Sequencing.

We deeply sequenced the cmDNA and nuclear genomic DNA (gDNA) isolated from WIL₂ cells. The 100-bp sequencing reads of cmDNA and gDNA were mapped to the entire human genome using Bowtie (12); 71,296,778 of 85,503,128 cmDNA reads (83.38%) and 70,642,258 of 82,711,889 gDNA reads (85.41%) were mapped to the genome. To identify regions of the genome enriched within the cmDNA dataset, we performed a tag-enrichment analysis to identify peaks of cmDNA that were enriched relative to gDNA. Only uniquely mapped reads were considered for this analysis, and 66,020,713 cmDNA reads (92.60% of mapped cmDNA reads) and 64,254,766 gDNA reads (90.96% of mapped gDNA reads) were studied. All the most prominent peaks were associated with satellite repeats of simple (CCATT)_N and 171-bp α-satellite repeat sequences (ALRs) or with genomic elements (subtelomere, ribosomal RNA) known to be associated with human satellite repeats (Fig.1A).

Fig. 1.

Deep sequencing of cmDNA. (A) Distribution in human chromosomes of DNA sequences of α-satellite repeat, pericentromeric, subtelomeric, and ribosomal RNA fragments that were highly enriched in cmDNA compared with gDNA. LSU-rRNA, large subunit ribosomal RNA; SSU-rRNA, small subunit ribosomal RNA. (B) Simple (CCATT)_N repeats and 171-bp ALRs were highly enriched in the cmDNA compared with gDNA. (C) The DNA sequence of the 171-bp ALR that was transcribed into mRNA in the WIL₂ cells.

In the human genome, satellite DNAs include small satellite repeats (satellites I, II, and III) located in pericentromeric regions and the 171-bp ALRs in centromeres (13, 14). To investigate whether the small satellite repeats are enriched within cmDNA, all our 100-bp reads were split into five 20mers per read, and the frequency of each unique 20mer, observed at least twice in the data, was compared in the cmDNA versus gDNA data. Of the top ten 20mers enriched in cmDNA versus gDNA, nine corresponded to simple satellite DNA (Fig. 1B).

We next investigated whether the 171-bp ALRs also are enriched in cmDNA relative to gDNA. We mapped reads using Bowtie to human repetitive elements cataloged in RepBase (15). This analysis confirmed that ALRs are highly enriched within cmDNA (Fig.1B). A breakdown of simple satellite repeats and ALRs and their frequencies in cmDNA versus gDNA are detailed in Datasets S1 and S2. Our deep-sequencing data are in agreement with previous studies and conclusively show that cmDNA is not a random sampling of the genome.

ALR DNA Is Transcribed into mRNA.

The deep sequencing indicated that there was a TATA box element sequence upstream of the 171-bp ALR gene. To examine whether this sequence was transcribed in vivo, total RNA isolated from WIL₂ cells was converted into cDNA that was used for PCR amplification of the 171-bp ALR fragment. Agarose gel electrophoresis and Sanger sequencing showed that the 171-bp ALR was amplified (Figs. 1C and 2 A and B). To ensure that the PCR products representing the 171-bp ALR were directly from the mRNA in the WIL₂ cells and not from cmDNA, total RNAs were treated with DNase I and/or RNase A, before conversion into cDNA and PCR amplification. The 171-bp ALR repeat could be amplified after DNase I treatment of isolated total RNA but not after RNase A treatment, confirming that the analyzed DNA product originated from mRNA (Fig. 2 A and B).

Fig. 2.

ALR mRNA is present in the cytoplasm and highly purified cytoplasmic membrane fragments prepared from WIL₂ cells. (A and B) To study whether ALR mRNA is present in the total RNA of WIL₂ cells, PCR amplification of ALR fragments from cDNA was attempted from fractions treated with DNase I (A) or with both DNase I and RNase A (B) before synthesis of cDNA and PCR amplification. (C and D) The purity of the purified cytoplasmic membrane fragments attached to the chitin magnetic beads was studied by fluorescent microscopy. (Magnification: 20×.) The EGFP protein was fused to a transmembrane domain, and a chitin-binding domain was expressed on outer surface of the cells. (C) EGFP is expressed on the cytoplasm side of the stable cell line WIL₂-CG. (D) The red color was developed with an antibody to pan cadherin, a plasma membrane marker. (E) Agarose gel electrophoresis of cmDNA isolated from the purified cytoplasmic membranes.

Special Transcription System.

We wondered whether cmDNA could be transcribed in the cytoplasm. To ensure that isolated plasma membrane fragments would be as pure as possible, we used a recently developed system for their isolation. In this system WIL₂ cells are transformed with a vector encoding a chitin-binding domain attached to a membrane-spanning domain and EGFP. Cells transformed with this vector express the chitin-binding domain on the outer surface of the plasma membrane and EGFP on the inner surface. Using this system, plasma membrane fragments can be highly purified using chitin magnetic beads. Analysis of membrane fragments attached to the chitin beads by fluorescent microscopy for the presence of membrane markers such as pan cadherin and EGFP showed that the attached membrane fragments originated from the plasma membrane (Fig. 2 C and D). Agarose gel electrophoresis of extracts of these membrane fragments showed that they contained cmDNA (Fig. 2E). The presence of ALR DNA in these purified membrane fragments was confirmed by demonstrating that the 171-bp ALR could be amplified by PCR from membranes that had been treated with RNase A (Fig. 3A). To rule out contamination, we showed that human β-actin (actin) and mitochondrial cytochrome B (cytoB) genes were not amplified (Fig. 3A). When cDNA was prepared from the DNase I-treated purified membrane fragments, PCR analysis showed that the 171-bp ALR was present, suggesting that, in addition to DNA, ALR mRNA is present in the purified membrane fragments (Fig. 3B). Finally, ALR, actin, and cytoB could not be amplified after the mRNA and DNA in the purified membrane fragments were digested by both RNase A and DNase (Fig. 3C). These results suggest that the purified plasma membrane fragments contained both ALR DNA and ALR mRNA and were not contaminated by nuclear and mitochondrial components.

Fig. 3.

PCR amplification showed that the purified plasma membrane fragments contained cmDNA, RNA polymerase II, and ALR mRNA. (A) To show that ALR could be amplified from membrane cmDNA, the purified and solubilized plasma membranes were digested by RNase A before PCR amplification of ALR, actin, and cytoB genes. (B) To show that ALR mRNA also was present in the purified membranes, the purified and solubilized plasma membranes were digested by DNase I before attempts to synthesize a cDNA template for PCR amplification of the ALR, actin, and cytoB genes. (C) Control shows no amplification of the ALR when the purified and solubilized plasma membranes were digested by both RNase A and DNase I before attempts to synthesize cDNA templates for PCR amplification of the ALR, actin, and cytoB genes. (D) Northern blotting of the newly synthesized RNA from in vitro transcription. To show that the membrane fragments were transcriptionally competent, RNA was synthesized by in vitro transcription of solubilized plasma membranes to which no exogenous template or RNA polymerase II was added. The newly synthesized RNA with biotin attached to UTP was detected with streptavidin-conjugated alkaline phosphatase The in vitro transcription negative control used the same reaction conditions except that GTP was not added to the reaction mixture. (E) To show that the amplified ALR originated only from newly synthesized mRNA from in vitro transcription, the purified and solubilized membrane fragments were digested by RNase A before in vitro transcription. After in vitro transcription, the solution was treated with DNase I to remove any DNA that could serve as a PCR template before cDNA synthesis. Lanes 1–4 show the purified and solubilized membrane fragments that were studied: 1, ALR; 2, actin; 3, cytoB; and 4, the PCR control to which no primers were added. Lanes 5–8 are the in vitro transcription negative controls in which the same reaction conditions were used but GTP was not added to the reaction mixture: 5, ALR; 6, actin; and 7, cytoB. In lane 8, as in lane 4, no PCR primers were added.

Because these studies showed that an ALR DNA template and its mRNA were present in the purified membrane fragments, it seemed possible that transcription could occur at this site. To study whether the purified membrane fragments were transcriptionally competent, they were added to an in vitro transcription system containing biotin-labeled UTP in which the only source of template DNA and RNA polymerase was from the membrane fragments themselves. Remarkably, as revealed by Northern blotting, an RNA of about 1.2 kb was synthesized by this in vitro transcription, even though no template DNA or RNA polymerase was added (Fig. 3D).

To confirm that the PCR-amplified ALR originated from newly synthesized mRNA prepared by in vitro transcription and not from preexisting mRNA or DNA, the purified membrane fragments were digested by RNase A before in vitro transcription and then were treated by DNase I after in vitro transcription to remove any DNA that could serve as a PCR template before cDNA synthesis (Fig. 3E). Study of these preparations by gel electrophoresis and Sanger sequencing showed that the 171-bp ALR, but not actin or cytoB, could be amplified from the cDNA transcript of the newly synthesized RNA prepared by in vitro transcription (Fig. 3E). The results suggested that RNA polymerase II was present in the purified membrane and could transcribe ALR DNA into mRNA.

Nature of Plasma Membrane-Associated RNA Polymerase.

We studied the properties of the DNA-dependent RNA polymerase bound to the purified membrane fragments by fluorescent microscopy and Western blot analysis, determination of its sensitivity to enzyme inhibitors, and analysis of its posttranslational modifications compared with nuclear RNA polymerases. Western blot analysis of the proteins present in the purified plasma membranes showed the presence of RNA polymerase II (Fig. 4 A and B). No RNA polymerase I or III was detected by Western blot analysis. The activity of the plasma membrane-associated RNA polymerase was inhibited by 1 μg/mL α-amanitin, confirming the presence of RNA polymerase II (Fig. 4C). Finally, fluorescence microscopic analysis of the purified cytoplasmic membrane fragments attached to the surface of chitin magnetic beads showed that RNA polymerase II was present Fig. 4 D and E).

Fig. 4.

Nature of the plasma membrane RNA polymerase II. (A and B) Silver staining (A) and Western blotting (B) of total proteins in the purified cytoplasmic membranes compared with nuclear extracts. The bands in Western blotting showed that the largest subunit of RNA polymerase II was present in both fractions. (C) The dot blotting of RNA synthesized from in vitro transcription in which the membrane-associated RNA polymerase II was inhibited by α-amanitin. The purified cytoplasmic membranes were treated with 1 μg/mL or 10 μg/mL α-amanitin. Also shown are the in vitro transcription negative control in which GTP was not added to the reaction mixture and the in vitro transcription positive control using the complete system without inhibitor. (D and E) Analysis of RNA polymerase II attached to the purified cytoplasmic membrane on the surface of chitin magnetic beads. (Magnification: 20×.) (D)The green color is EGFP. (E) The red color is anti-human RNA polymerase II antibody developed with a secondary antibody labeled with DyLight 550. (F) Western blotting of Ser5 phosphorylation in the CTD of the largest subunit of RNA polymerase II from the purified plasma membrane compared with isolated nuclear extracts. Phosphorylation of Ser5 in the RNA polymerase II CTD was present in nuclear extracts but not in the purified plasma membrane. (G) Western blotting showed that nearly equal amounts of RNA polymerase II from the nuclear and plasma membrane extracts were loaded onto the gels. Irrelevant primary antibodies were used as negative controls in the Western blots.

To determine if there were differences between the nuclear and plasma membrane-associated RNA polymerase II, we studied the phosphorylation state of the C-terminal domain (CTD) of the largest subunit of RNA polymerase II. RNA polymerase II occurs mainly in the cell nuclei, transcribes DNA into mRNA and noncoding RNAs, and has properties different from those of RNA polymerase I and RNA polymerase III (16, 17). Phosphorylation of the CTD in RNA polymerase II is associated with the progress of transcription and mRNA processing (18, 19). Our studies showed that Ser5 was phosphorylated in RNA polymerase II isolated from the nucleus but was not phosphorylated in the RNA polymerase II attached to the plasma membrane (Fig. 4 F and G). This result suggested that the membrane RNA polymerase II and nuclear RNA polymerase II have different properties, at least in regards to Ser5 phosphorylation. Phosphorylation of Ser2 or Thr4 was not detected for either nuclear or plasma membrane-associated RNA polymerase II.

Discussion

In these studies we took great care to ensure the purity of the cell fractions studied. Although no cell-fractionation procedure can be expected to give absolutely pure material, our system using the expressed chitin-binding domain provided plasma membrane preparations of very high purity. Because the results of our study relied on enzymatic amplification, the proper and most stringent control is the study of a molecule equally capable of being amplified. In this regard, the high degree of purity of our preparations was confirmed in the in vitro transcription studies, which showed that, unlike ALR DNA, no actin or cytoB sequences were transcribed. Also, studies of polymerases showed that only one (RNA polymerase II) of the three possible polymerases was associated with the plasma membrane, and that polymerase appeared to be in a different chemical state from its nuclear counterpart.

The studies presented here may change the dogma that, with the exception of mitochondrial genes, DNA transcription occurs in the nucleus. Although the role of transcription at plasma membranes remains obscure, several models can be proposed. One model assumes that transcription at the plasma membrane is involved in signal transduction, so that the newly formed RNA or its derivatives are transported to the nucleus where they can modulate gene expression. A more conservative suggestion is based on previous experimental data concerning the role of transcripts of α-satellite DNA in the formation of the chromosomal centromere and the need to resolve the dilemma posed by proposing a role for transcripts that emanate from DNA that is largely transcriptionally silent. The most thorough study of the role of centromeric satellite DNA in the formation and integrity of the centromere was carried out in beetles (20); this study showed that the satellite transcripts migrate to the nucleolus together with centromeric DNA-binding proteins that also have affinity for the centromeric RNA transcripts. Then, during prophase, the RNA attaches via the binding proteins to the α-satellite repeats in the DNA, where the complex plays a critical role in heterochromatin formation and establishment of the kinetochore. Thus, one can propose a model in which the α-satellite DNA is amplified, perhaps by “onion skin” replication, after which RNA polymerase attaches and the complex migrates to the plasma membrane where transcription occurs. Because the complex at the plasma membrane is transcriptionally competent, the problem of generating RNA from otherwise silent regions of the genome is solved. In accord with current thinking, such transcripts play a role in gene silencing and maintenance of chromosomal integrity (21, 22). Finally, it has been shown in a variety of species that satellite transcripts are involved in the RNA interference pathway that affects histone modification (20).

Although we would like to consider a normal physiological function for cytoplasmic DNA and its transcripts, two recent studies suggest that it may have a role in oncogenesis. Ting et al. (21) showed that in 15 of 15 human pancreatic tumors satellite DNA transcripts showed a mean 21-fold increase in expression compared with normal tissues. In particular, the α-satellite component of repeated DNA showed a 43-fold overexpression in tumor tissue. Zhu et al. showed that loss of the tumor-suppressor gene BRCA-1 resulted in de-repression of α-satellite DNA (22). The present study may add to these important results by indicating that the overexpression of α-satellite DNA in cancer might result from inappropriate cytoplasmic transcription.

Materials and Methods

Cell Culture and Isolation of cmDNA and gDNA.

The human diploid B-lymphocyte cell line WIL₂ was purchased from the American Type Culture Collection. A stable WIL₂-CG cell line, which expressed a chitin-binding domain on the cell surface, was constructed in our laboratory. The cmDNA and gDNA were isolated as described previously. See SI Materials and Methods for further details.

Deep Sequencing and Bioinformatics of Sequence Analysis.

cmDNA and gDNA were deeply sequenced with an Illumina GA-II sequencer, and 100-bp reads of cmDNA and gDNA were mapped to the entire human genome using Bowtie and human repetitive elements cataloged in RepBase. See SI Materials and Methods for further details.

Total RNA Isolation, cDNA Synthesis, and ALR Fragment Amplification.

Total RNA was isolated from WIL₂ cells using the RNeasy midi kit (Qiagen). cDNA was generated using SuperScript III First-Strand Synthesis SuperMix (Invitrogen) and oligo (dT)₂₀ primers. The amplification of ALR from cDNA was conducted using the HotStarTaq Master Mix Kit (Qiagen). Total RNAs were treated by DNase I (Ambion) or by both RNase A (Ambion) and DNase I (Ambion). The treated solutions were reverse transcribed into cDNA, and ALR repeats were amplified by PCR. See SI Materials and Methods for further details.

Purification of Cytoplasmic Membrane and in Vitro Transcription.

A stable cell line, WIL₂-CG, was used to purify the cytoplasmic membrane. WIL₂-CG cells were lysed using reticulocyte standard buffer with 0.03% Nonidet P-40 and protease inhibitor mixture (Sigma) for 10 min at 4 °C. The lysates were centrifuged at 1,000 × g at 4 °C for 5 min to sediment nuclei. The membrane fragments in the supernatant solution were bound to chitin magnetic beads (BioLabs). The nuclei in the pellet were dissolved with 1% Nonidet P-40, and DNA was isolated. cmDNA was isolated from the purified cytoplasmic membrane as described above. The purity of the cytoplasmic membrane fragments attached to the chitin beads was validated by fluorescent microscopy. In vitro transcription was conducted using the AmpliScribe T7-Flash Biotin-RNA transcription kit (Epicentre, Illumina). See SI Materials and Methods for further details.

Northern Blotting.

The sample from in vitro transcription was separated on 1% agarose formaldehyde-free RNA gels using the positively charged nylon membranes and BrightStar BioDetect Kit for nonisotopic detection of biotinylated RNA (Ambion/Invitrogen). See SI Materials and Methods for further details.

PCR Amplification of ALR, Actin, and cytoB Fragments.

The purified and solubilized membrane fragments were treated with RNase A, DNase I, or both, and then PCR amplifications of ALR, actin, and cytoB fragments were conducted using the treated membranes as templates. The purified and solubilized membrane fragments were digested by RNase A before in vitro transcription. After in vitro transcription, the membrane fragments were treated with DNase I to remove any DNA that could serve as a PCR template before cDNA synthesis. Then cDNA synthesizing and PCR amplification of the ALR, actin and cytoB fragments were conducted. See SI Materials and Methods for further details.

Western Blotting of Proteins in the Purified Membrane.

The primary antibody was at a final concentration of 1.0 μg/mL The secondary antibody conjugated with HRP was at a final concentration of 100 ng/mL SuperSignal West Pico Chemiluminescent Substrate (Thermo Scientific) was used for detection and visualization. See SI Materials and Methods for further details.

Studies of RNA Polymerase II.

The purified cytoplasmic membrane fragments bound to chitin magnetic beads were studied. The primary antibody was mouse anti-human RNA polymerase II (Novus Biologicals) in a final concentration 10 μg/mL. The secondary antibody was goat anti-mouse IgG (heavy and light chain) conjugated with DyLight 550 (Thermo Scientific) at a final concentration 10 μg/mL See SI Materials and Methods for further details. The purified cytoplasmic membranes were treated with α-amanitin (Sigma) at a final concentration of 1 μg/mL or 10 μg/mL In vitro transcription reactions were conducted as described above and were analyzed by dot blotting. See SI Materials and Methods for further details.